Across-species differences in pitch perception are consistent with differences in cochlear filtering

Abstract

Pitch perception is critical for recognizing speech, music and animal vocalizations, but its neurobiological basis remains unsettled, in part because of divergent results across species. We investigated whether species-specific differences exist in the cues used to perceive pitch and whether these can be accounted for by differences in the auditory periphery. Ferrets accurately generalized pitch discriminations to untrained stimuli whenever temporal envelope cues were robust in the probe sounds, but not when resolved harmonics were the main available cue. By contrast, human listeners exhibited the opposite pattern of results on an analogous task, consistent with previous studies. Simulated cochlear responses in the two species suggest that differences in the relative salience of the two pitch cues can be attributed to differences in cochlear filter bandwidths. The results support the view that cross-species variation in pitch perception reflects the constraints of estimating a sound's fundamental frequency given species-specific cochlear tuning.

Keywords: cochlea; ferret; hearing; human; neuroscience; pitch; psychophysics; resolved harmonics.

Figure 1.. Simulated cochlear filters and their responses to a 500 Hz harmonic complex tone filtered from 1 to 10 kHz.

(A) Illustration of the role of unresolved and resolved harmonics in periodicity encoding. Upper plot: Amplitude spectrum for the tone complex contains all harmonics from 1 to 10 kHz. This sound will evoke a pitch corresponding to 500 Hz. Middle plot: Cartoon of the cochlear filters centred on every second harmonic of 500 Hz, based on data from Glasberg and Moore (1990). This illustrates that lower harmonics are resolved, while the cochlear filters corresponding to higher order harmonics respond to multiple harmonic components in the tone. Lower plot: The output of each of these cochlear filters is plotted throughout 5 ms of the tone complex. The resolved harmonics phase lock to the frequency of one harmonic, while unresolved harmonics beat at the sum of multiple harmonic components (i.e. 500 Hz), providing an explicit temporal representation of F0. B-E describe a computational model of the cochlear filter bank used to simulate representations of complex sounds in the ferret and human auditory nerve. Data are colour-coded for the human (blue) and ferret (black). (B) The frequency tuning of 15 example auditory nerve fibres is shown for the simulated human (left) and ferret (right) cochlea. C-E show analyses of the responses of human and ferret cochlear filter banks to the 500 Hz tone complex. (C) The response strengths of each of 500 auditory nerve fibres to the filtered tone complex were averaged across the duration of the sound, and plotted across the full range of centre frequencies. Many harmonics produce clearly resolvable activation peaks across fibres in the human cochlea (upper blue plot), but fewer harmonics are resolved in the ferret cochlea (lower black plot). (D) The temporal profile of the output of one simulated auditory nerve fibre with a centre frequency of 5 kHz is shown for the human (upper blue plot) and ferret (lower black plot) cochlea. (E) The power at 500 Hz in the output of each frequency filter, averaged across the full duration of the tone complex, is shown for the human (blue) and ferret (black) auditory nerve. For each species, these values were normalized by the maximal F0 power across all channels. The plot shows the mean (+standard error) normalized power at F0 across all auditory nerve fibres.

Figure 2.. Psychophysical task design.

(A) Schematic of the ferret testing apparatus, viewed from above. (B) Schematic of one trial in the 2-alternative forced choice pitch classification task. The target tone complex could be lower or higher in F0 than the reference pure tone (R). Dotted lines indicate time durations that are variable, depending on the animal’s behaviour.

Figure 3.. Stimuli used in the ferret pitch classification task.

Plots show the training tone (left column), standard stimulus (second column) and four probe stimuli (columns 3–6) used in the psychophysical task. There were two F0 ranges, with target stimuli with F0s of either 150 and 450 Hz, or 500 and 1000 Hz, indicated to the left of each row of plots. The top four rows show the power spectra of each target sound, while the bottom four rows plot a 20 ms excerpt of the corresponding sound waveform. The table in the middle of the figure indicates whether resolved harmonics (row 1) or temporal envelope (row 2) F0 cues are strong or weak in each stimulus. A pure tone reference sound was used in all experimental stages.

Figure 4.. Harmonic content of stimuli.

(A) The number of resolved harmonics was estimated over a range of F0s, for the human (blue) and ferret (black) cochlea. (B) The frequency ranges and numbers of harmonic partials included in each stimulus. As per convention, F0 is deemed to be harmonic 1.

Figure 5.. Pitch classification performance of ferrets and humans.

(A) Ferrets’ percent correct scores on the pitch classification task are plotted for the standard tone trials (left) and each of the four probe stimuli (right). The results of testing with the 260 Hz reference (150 and 450 Hz targets; red) and 707 Hz reference (500 and 1000 Hz targets; black) are plotted separately. (B) Humans’ pitch classification performance is plotted, as in (A). (C) Performance for each of the four probe stimuli is expressed as the ratio of the percentage correct score and that achieved with the standard training tone stimulus. Data are shown for ferrets (black) and humans (blue). Values of 0 indicate that subjects performed at chance for the probe stimulus, while one indicates that they classified the F0 of the probe as accurately as the standard stimulus. Error bars shown mean ± standard error. Individual data for (A) and (B) are shown in Figure 5—figure supplement 1.

Figure 5—figure supplement 1.. Pitch classification performance of individual ferrets and humans, as shown in Figure 5A and B.

(A) Ferrets’ percent correct scores on the pitch classification task are plotted for the standard tone trials (left) and each of the four probe stimuli (right). The results of testing with the 260 Hz reference (150 and 450 Hz targets; red) and 707 Hz reference (500 and 1000 Hz targets; black) are plotted separately. Symbol shapes represent individual ferrets. (B) Humans’ pitch classification performance is plotted, as in (A). Data are randomly jittered on the x-axis to facilitate visualization of individual points. Each symbol and colour combination indicates an individual subject.